Estimating Renal Function for Drug Dosing: Rewriting the Gospel? part 3
THE NEW TESTAMENT VERSUS THE OLD TESTAMENT
The National Kidney Foundation has recently reviewed the literature related to the predictive performance of these equations for GFR, as measured by iothalamate or iohexol clearance. For the CG equation, the proportion of estimates within 30% of actual (measured) GFR ranged from 48% to 95%. The full MDRD equation yielded better predictions, with 88% to 91% of GFR estimates being within 30% of measured values. The abbreviated MDRD equation performed almost as well as the full MDRD equation, with 84% to 91% of GFR estimates within 30% of measured values. One of the reasons that the MDRD equations perform better than the CG equation in these situations is because they were designed to estimate GFR, whereas the CG equation was designed to estimate creatinine clearance.
Recently, Wargo and others found that rates of antimicrobial dosage discordance (differing dose recommendations) between doses based on the CG equation and those based on the MDRD equations ranged from 21% to 37%. The majority (86%) of the discordances occurred when the estimate from the CG equation dictated a dosage adjustment but the estimate from the MDRD equations did not. The authors stated that the patients would have been “overdosed” 21% of the time if the MDRD equation had been used.16 However, because the authors did not look at any clinical parameters to assess proper dosing, it is probably more appropriate to simply state that the doses would have been different. In addition, the assumption that the dose determined by the CG equation is correct is just that: an assumption. There is little if any evidence showing that doses calculated this way are clinically correct.
canada viagra online
de Lemos and others examined the impact of using the abbreviated MDRD and CG equations in calculating carboplatin doses with the Calvert equation (Equation 6 in Appendix 1). For all patients, actual GFR was measured with diethylenetriaminepenta-acetic acid labelled with technitium-99m for comparison. Use of either of the 2 equations to estimate GFR would have led to lower doses than if the measured GFR values had been used (doses of 622 mg by measured GFR, 557 mg by the CG equation, and 575 mg by the abbreviated MDRD equation). There was no statistically significant difference in GFR estimates between the CG and abbreviated MDRD equations (p = 0.68). Even if the 18-mg (3%) difference in calculated dose between the 2 equations was real, it is unlikely that it would produce a clinically important difference. The discrepancy between the doses calculated with measured and estimated GFR is not surprising, as the Calvert equation causes the dose of carboplatin to change in direct proportion to renal function, whereas most drug dosing tables have broad categories. Although de Lemos and others used the Calvert equation as the reference equation, there was no way to determine the “correct” dose for these patients.
Most recently, Gill and others examined differences between GFR estimates for elderly patients in long-term care centres and described the effect of these estimates on dosing of 2 medications, amantadine and digoxin. For 180 patients, the mean GFR estimates were 72.9 mL min-1 1.73 m-2 with the abbreviated MDRD equation and 52.1 mL min-1 1.73 m-2 with the CG equation. With the abbreviated MDRD equation, 21.2% fewer patients would have received an amantadine dosage reduction and 32.2% fewer patients would have received a digoxin dosage reduction. The authors acknowledged that the “correct” dose was undetermined and recommended “caution” when using these formulas for drug dosing.